Google searches bad. What to do?

A few days ago, I asked my Twitter readers if they thought Google searches seemed worse lately.

Almost 73% (41.4+31.5) agree that Google has been getting worse at searching in recent years.

But I think if I polled random people on the street, I think about only 10–20 per cent would agree. In my Twitter account

I talk about tools and techniques for OSINT (Open Source Intelligence) and I am read by a lot of people who are professionally engaged in searching for information on the Internet.

Google has actually gotten better at simple queries like “Paul Altman’s age” (thanks to “quick answers”), but ut don’t good for complex search tasks, like finding the maximum number of mentions of a certain non-media person or seeking answers to complex scientific questions.

Good old days legends

Unfortunately, I can’t pinpoint the exact time frame of the period when Google searches did really well. My feeling is that in 2008–2013 most of the search results matched the query and Google was a pleasure to use. But even then, some of the problems described below were already occurring.

2014–2019 were also pretty good years (but the problems have become more and more frequent). Last few years something really wrong has been going on. This became especially noticeable from winter 2022–2023, when there was too much AI-generated text in the results.

But… This article is most likely far from the first text about Google getting bad that you read. After all, they have been talking about it for at least 20 years (and more likely, more than 25, because the search engine appeared in 1998).

But the main purpose of this text is not to determine when Google started to search badly, but to specify the problems and suggest possible solutions.

Google search problems

The quality of results has become lower

Lately, Google has started displaying more and more results that are completely irrelevant to the search query (or that are irrelevant based on the snippet of text that appears on the main results page)

If the query is highly specialised and there are few results for it, there may be only 1–2 relevant results for the all results.

Google no longer understands long and complex queries

On Quora, you’ll sometimes see examples of search queries from users that used to work for them and then stopped. For example, “where is 1978 Oldsmobile windshield washer pump timer circuit” was finding the answer a couple of years ago, but now it’s like Google has basically stopped understanding it.

The more words there are in a query, the worse Google performs. Although I think 10 years ago long queries worked much better.

The impact of filter bubble has increased

Information about long snakes has become quite impossible to find. In addition, thanks to Google, I clearly saw that there is a JavaScript library for almost every word.

The results depend not only on the current query, but also on all the queries you have entered since the account was created.

The impact of the user’s location has also become much more powerful. I recently came across the fact that even if I add the name of the city to the name of the hairdressing salon, the results still show hairdressing salons that are near me, but not in the corresponding city.

Limitation on the number of results

In the mid-2000s, it was realistic to go to the 50th or 100th page of search results and find something there. Now it is very rare to see more than 100–200 results.

And with that, the last 50–100 results can match the query very poorly.

Censorship, copyrights, privacy etc.

More and more results are being hidden from searches from the fact that the text pages contain infringing information. More and more texts are hidden from searches due to copyright infringements (links to it can be found in the Lumen Database, so this is pointless and uncomfortable).

And, the saddest thing for OSINT specialists, personal data has also started to be hidden from search results (3–4 years ago, search by email was much more efficient).

SEO advertising

The only thing worse than SEO texts, as it turns out, can be AI-generated SEO texts.

A huge number of texts written only for the purpose of selling something and containing false information are unfairly placed at the top of the search results.

Sometimes this is combined with 3–5 Adsense adverts on one page and is absolutely frustrating.

What to do?

Unfortunately, I don’t have a short answer to that question. I can’t just say “use Bing” or “use Baidu” as all modern search engines are far from perfect and Google, if used correctly, is still one of the best.

So the main advice is “Just because Google suddenly didn’t find something doesn’t mean it’s not exist. Try different search methods and tools”.

Search filters and more precise query formulation

The most effective ways, in my opinion, to improve Google search results would be a combination of using quotes and the intitle: operator (search by page title)

intitle:”osint collection”

Quotes (“exact match search”) can be used without search operators. Although it seems that they have recently become worse (but this may be due to the fact that the text of some sites changes after indexing and the search is also based on hidden text, like alt attributes of pictures).

If you want to see less text written by AI and SEO specialists and more text written by live people, then use the site: operator to search for social media posts.

site:reddit.com

site:facebook.com

If you need to search for something related to science or medicine, search only on research papers:

site:researchgate.net

site:academia.edu

And, of course, don’t forget about other Google search operators.

Also, try to choose synonyms as carefully as possible and make fewer mistakes in your query. Google used to be good at understanding what the user “really meant”. Now this functionality has become worse.

Location manipulation

It can be very helpful to step out of the filter bubble to improve the quality of your search results. For start, you can simply use incognito mode in your browser and some service to change your location and language.

isearchfrom.com

brightlocal.com

Other search engines

I think many of you have tried switching to DuckDuckGo or another private search engine, but ended up going back to Google anyway.

As far as other search engines are concerned, I can only highly recommend only to also use Bing and Yandex to search for pictures.

In cases where Google does not give results in text search, try different search engines that you like, hoping to get lucky with some of them. If you know in which country the site you need is located, try to use national search engines (sometimes it brings very good results).

AI assistants

Lately, more and more people have been abandoning Google in favour of AI assistants. Here are the three main things I can say about them:

1. They shamelessly lie.

2. ChatGPT is not always the best choice. Perplixity.ai or you.com are pretty good too (they’re at least faster and free to search the web for information). But they are also shamelessly lie.

3. Browser extensions that add AI assistant responses to Google search results is a real handy thing. For example, MaxAI allows you to access different AI assistants from any browser tab and visually see that they all shamelessly lie differently.

AI assistants for searching through research papers are also worth mentioning. They provide answers strictly based on scientific publications (with references to specific publications).

Elicit.com

ResearchAIDE.com

Sometimes I think they miss some researches, but in general they are much more truthful than other AI assistants (but they still shamelessly lie sometimes).

IP search engines

IP search engines like Netlas, Shodan, Censys, Fofa crawls all the IP addresses on the internet and are good because when you search by word they:

- NOT range
- NOT filtering
- NOT distort search queries

they just look for an exact match on all the sites and servers.

For example, in Netlas you can search by web page body:

http.body:keyword

Web page title:

http.title:keyword

Web page meta tags:

http.meta:keyword

And filter results by many different parameters (domain, url, geolocation etc).

But the best part is that you can not only see all the search results up to the very last page, but you can also DOWNLOAD them to your computer (full html codes of the pages) to do deeper searches locally.

(Uploading hundreds of thousands of results can take a long time, so it’s best not to upload more than a few thousand unnecessarily).

The biggest disadvantage of IP search engines is that they only search the main pages of websites. After all, they add to their database only server responses from different IP addresses and do not bypass pages link by link, as it is done by conventional search engines.

Therefore, they cannot be a full-fledged replacement for Google. Only an additional tool.

Crawlers, scrapers and local search tools

A good option would be to search the main pages of sites using Netlas (for example, you can collect all sites that have a certain word in their title) and then organise a local search on your computer or server.

To do this, you need to first get a list of pages on the sites, using some kind of scraper, such as Scrapy Python Package, then load them with Curl, and then use Grep or analogues to locally search the text of html files.

You can also use archive.org to search for website pages, use tools to automate the collection of search results such as duckduckgo-search or do bruteforce links using a dictionary of popular filenames using GoBuster. Read more about this here:

8 basic methods of automating the collection of information from company websites

Archives and libraries search

Before you search for something using universal search tools, consider whether it is better to use thematic archives, libraries, forums. The more highly specialised a site is, the more likely you are to find something of real value on it.

However, universal libraries and archives are also sometimes more convenient than Google. I would especially like to recommend Anna’s Archive.

Conclusion

In this article, I have tried to put together as many universal tips as possible. But the main rule of effective search is to approach each task as individually and creatively as possible.

On my Twitter Cyber Detective, I regularly write about different methods and tools for searching for information on the Internet.

Due to lack of time, don’t miss out on any helpful info about the OSINT community. Read our free weekly newsletter and stay informed about the OSINT community for free. Join the free newsletter at osintupdates.com right now.

Stop wasting time reading the same content about OSINT again and again. Read high-quality, non-repeated OSINT blogs for free at osintblogs.com! Don’t forget to subscribe so you don’t miss out on new blogs.